Study On Configuration Synthesis Of Reconfigurable Robots Based On Fault Tolerant Indices

The reconfigurable robotic system is an intelligent robotic system that iscomposed of a series of standard modules with various sizes. According to therequirements of different working environments and tasks, it can easily and rapidlychange its mechanical structure to build the best operating configuration. Obviously,configuration synthesis is a key issue of the reconfigurable robot research. Theredundant robotic system, emerging and developing from the late 1980s, is also anadvanced robotic system. The research that focuses on how to make the operation ofthis kind of robot more available for fault tolerance by means of the redundancycharacteristic of the robot's own configuration has been one of the most active fieldsin the redundant robot research. By introducing the concept of redundancy into thereconfigurable robotic system and combining the advanced characteristics of thesetwo kinds of robot mentioned above, the reconfigurable robotic system will have anoutstanding superiority and will find wide applications infault tolerant operations. Inthis dissertation, the problems onthe fault tolerance planning ofthe redundant robot,the conceptual design and automatic kinematical modeling method of thereconfigurable robotic system, and the configuration synthesis of the redundantreconfigurablerobotbasedonfaulttolerantindiceshavebeen studiedelaborately.Firstly, thebasictheoryandthe kinematicsoptimizationmethodoftheredundantrobot are introduced. The two kinds of fault tolerant indices, the fault tolerantworkspace and the reduced condition number, are theoreticallyanalyzed and properlymodified. Considering both of the reduced condition number and the joint motionlimit criterion, the motion planning optimization index function is constructed. Thisindex function can not only make the robot possess a high flexibility in fault tolerantoperation, but also ensure that each of the robot's joints would run within its jointmotionlimitatthesametime.Secondly, according to the application requirements, the general designprinciples of a reconfigurable robotic system are summed up. Based on theseproposedprinciples,theconceptual designofthereconfigurableroboticsysteminthisdissertationis done. Then, theautomatickinematical modeling methods of theroboticsystem are studied, including the methods to solve the robot's forward kinematicequation and its Jacobian matrix. These auto-modeling methods overcome theinconvenience brought by the traditional D-H method as the D-H method has toderive the corresponding kinematic model respectively according to differentconfigurationsoftherobots. Furthermore, the problem on configuration synthesis of the reconfigurable robotbased on the fault tolerant indices is explicitly described. Then, the fundamentaltheory of the genetic algorithm that is used to deal with the configuration synthesisproblem is introduced and explained. The solving process of the configurationsynthesis problem in this dissertation is divided into two stages, the configurationsearching stage and the configuration selectionstage. The methods and procedures ofthe implementation algorithms in these two stages are elaborated, in which the detailsof the genetic algorithm that is used toimplement the configuration searchingstage isemphatically expounded, including the methods to encode and decode theconfigurationchromosome,thewaytogeneratetheinitial population,andthespecificmethods to run those basic genetic operators, such as selection, crossover andmutation.Finally, by utilizing Matlab the configuration synthesis algorithm proposed bythis dissertation is carried out. For some given tasks, configuration synthesissimulationexamplesforaplanar3DOFandaspacial4DOFreconfigurablerobotareaccomplished based on the fault tolerant indices. The simulation results show thattheconfiguration synthesis algorithm proposed by this dissertation can effectively findthe configuration of the reconfigurable robot which possesses the relatively optimalfaulttolerantperformance.